ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Robot "bucking" while driving in reverse

asked 2018-08-09 20:02:26 -0500

pitosalas gravatar image

updated 2018-08-14 05:17:24 -0500

Doing the Slam experiments, my robot does this (see attachment) buckling or stuttering motion. What causes that?

image description

edit retag flag offensive close merge delete



@pitosalas: I took the liberty of converting your video to a gif so we keep this question self-contained.

gvdhoorn gravatar image gvdhoorn  ( 2018-08-10 02:33:11 -0500 )edit

Thanks. I couldn’t find a way to attach a video. Animated gif. Clever 😀

pitosalas gravatar image pitosalas  ( 2018-08-10 14:39:14 -0500 )edit

1 Answer

Sort by » oldest newest most voted

answered 2018-08-09 20:43:47 -0500

Are you sending two 'cmd_vel' topics? If so, one might be sent correctly and one seems to be at zero speed. Check the 'cmd_vel' topic through rqt_graph.

edit flag offensive delete link more



A typical case for this to happen is when there is still some teleop node running somewhere when starting autonomous navigation.

Felix Widmaier gravatar image Felix Widmaier  ( 2018-08-10 01:51:42 -0500 )edit

@Pyo I suspected this might be the case but didn’t immediately think of how to check. Thanks for tip. @felixwidmayer I suspect the same. I think I checked that I killed it but there might have been a zombie teleport process that was not fully killed.

pitosalas gravatar image pitosalas  ( 2018-08-10 14:41:57 -0500 )edit

@Pyo @felix-widmaier - I see "bucking" behavior when teleoping in reverse, with no other nodes or packages running. So the theory was good, it doesn't; really seem to apply. Rosnode list, shows this: rosout, rqt_gui_py_node_4440, turtlebot3_core, turtlebot3_lds, turtlebot3_diagnostics.Thoughts??

pitosalas gravatar image pitosalas  ( 2018-08-13 13:11:37 -0500 )edit

Perhaps a screenshot of rqt_graph would be more informative. Be sure to create it when you're actually observing the behaviour in the video.

It's very likely that there is a node publishing zero-Twists.

gvdhoorn gravatar image gvdhoorn  ( 2018-08-14 05:51:28 -0500 )edit

I will try but I am almost 100% sure because I looked at the graph and saw no-one publishing on cmd_vel.

I am suspecting that it is a weight distribution problem because it seems to change if I slide the battery around a little. It doesn't solve it but it seems more or less severe

pitosalas gravatar image pitosalas  ( 2018-08-14 10:28:01 -0500 )edit

Could also be there is no proper ramp-up of teleop velocity commands. I already commented on your #q300208 about how the TB2 addresses that.

gvdhoorn gravatar image gvdhoorn  ( 2018-08-14 10:29:16 -0500 )edit

@gvdhoorn by all means you've been super helpful and I am grateful! The teleop code supplied (and I looked at the source) definitely does do a rampup. I assume it's correct but I can investigate to make sure it does what is stated.

pitosalas gravatar image pitosalas  ( 2018-08-14 10:34:22 -0500 )edit

re: investigate: that may be a bit much at this stage.

I would take a look at some of the msgs on the /cmd_vel topic (or whatever is being used on the TB3). Just use rostopic echo /cmd_vel when the robot is standing still (replace with appropriate topic). Then compare when teleopping.

gvdhoorn gravatar image gvdhoorn  ( 2018-08-14 10:48:15 -0500 )edit

Question Tools

1 follower


Asked: 2018-08-09 20:02:26 -0500

Seen: 272 times

Last updated: Aug 14 '18